A face feature extraction method with illumination robustness

A technology of illumination robustness and facial features, applied in the directions of instruments, character and pattern recognition, computer parts, etc., it can solve the problems of complex models such as reflection, slow processing speed, and inflexibility, so as to simplify the calculation. with the effect of processing

Inactive Publication Date: 2009-09-23
SOUTHWEST JIAOTONG UNIV
View PDF4 Cites 3 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

However, it is more complicated to look for models such as light generation and reflection, and the requirements for equipment are relatively high, such as infrared devices, so this method is not flexible and the cost is relatively high.
Therefore, the second type of method is usually used, but the 3D algorithm complexity is relatively high, and the processing speed is relatively slow, which does not meet the requirements of real-time applications.

Method used

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View more

Image

Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
  • A face feature extraction method with illumination robustness
  • A face feature extraction method with illumination robustness
  • A face feature extraction method with illumination robustness

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0034] figure 1 As shown, the first specific implementation of the present invention is: an illumination robust face feature extraction method based on discrete Fourier transform phase reconstruction, including the following steps a to e:

[0035] a. Preprocessing of face images:

[0036] For the original two-dimensional face image I with the size of M rows and N columns P Apply the discrete Fourier transform to reconstruct the phase information only to obtain a binarized face preprocessing image with a size of M rows and N columns The specific method is as follows:

[0037] For the original two-dimensional face image I with the size of M rows and N columns P Do discrete Fourier transform, in this case M=N=128 promptly to the original two-dimensional face image I of 128 rows and 128 column sizes P . The formula of the one-dimensional discrete Fourier transform is:

[0038] y k = Σ i ...

Embodiment 2

[0067] This example is basically the same as Example 1, the only difference is that in step d, n global face feature vectors are obtained (P=1, 2, 3,...n), first use linear discriminant analysis (LDA) to perform dimension reduction processing, and obtain n global face feature vectors after dimension reduction (P=1,2,3,...n), construct and form face feature database again; The Euclidean distance of corresponding step e ED ( I fea 0 , I fea P ) = Σ q = 1 2 L ( EN q 0 - EN q ...

Embodiment 3

[0080] The third specific implementation of the present invention is: an illumination robust face feature extraction method based on edge information, which is basically the same as the first embodiment, except that the preprocessing of the face image in step a is different. The method is: for the original two-dimensional face image I of the size of M rows and N columns P Use the edge detection algorithm to extract the contour features of the face, and then perform binarization on the face contour image to obtain a face preprocessing image with a size of M rows and N columns after binarization In this example, the original two-dimensional face image I P The sobel edge detection algorithm is used to extract the contour features of the face. Its more specific operation instructions are as follows:

[0081] 1. Preprocessing of face images (extracting edge information):

[0082] 1. For the original two-dimensional face image I with M rows and N columns P Apply the sobel opera...

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to view more

PUM

No PUM Login to view more

Abstract

A face feature extraction method with illumination robustness, the steps of which are: use edge detection or discrete Fourier transform to preprocess the face image, divide the preprocessed face image into uniform blocks without overlapping, and calculate The row and column variance projection entropy of each block is combined with the extracted row and column variance projection entropy of each block into a face feature vector to construct a feature database of multiple face images. During identification and authentication, according to the same preprocessing and face feature vector extraction method, the face feature vector to be recognized is calculated, and the distance classification is carried out with the face image feature vector in the feature database one by one to obtain the recognition result. Its advantages are: there is no need to use complex lighting models or imaging equipment; it has good robustness to lighting, and it can also effectively extract the features required for face recognition from face images formed under poor lighting conditions. The recognition rate is high; and the processing speed is fast, with real-time processing capability.

Description

technical field [0001] The invention relates to a biometric feature automatic recognition technology, in particular to a feature extraction method for face recognition under changing illumination conditions. Background technique [0002] In recent years, with the rapid development of biometric recognition technology, face recognition technology has attracted more and more attention from researchers. This technology has huge application advantages in video surveillance, human-computer interaction, and identity authentication. At present, there are many studies on face recognition under controlled conditions, and the results are relatively good. However, under illumination changes, the effect of face recognition drops sharply, and this problem has not been effectively solved yet. Generally, there are several processing methods for face recognition under lighting conditions: [0003] (1) Physical preprocessing of light [0004] Since there are many physical mathematical mod...

Claims

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to view more

Application Information

Patent Timeline
no application Login to view more
Patent Type & Authority Patents(China)
IPC IPC(8): G06K9/00
Inventor 张家树陈存建
Owner SOUTHWEST JIAOTONG UNIV
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Try Eureka
PatSnap group products